The present invention relates to partitioned dataset extended (PDSE) critical dataset redundancy (CDR), and more specifically, this invention relates to methods for recovering from both in-core and on-media corruption with minimal interruption to operations.
Currently, PDSE processing is split between two address spaces, SMSPDSE which handles global PDSE connections, and SMSPDSE1 which handles local PDSE connections. Although SMSPDSE1 is able to be restarted in order to recover from errors and/or corruption to datasets, SMSPDSE is not because of multiple virtual storage (MVS) restrictions. Therefore, a globally connected PDSE currently cannot recover from corruption or damage without a system initial program load (IPL).
When a globally connected PDSE becomes corrupted, especially a critical dataset such as one in LINKLIB or a STEPLIB, critical jobs can fail leading quickly to the need for an emergency IPL. An emergency IPL causes a system outage and can result in millions of dollars in expenses and missed service level agreements, which clients typically look to be reimbursed for. While significant code improvements have been made over the life of PDSE, index-breaking errors are still prevalent and reducing or eliminating these errors would greatly benefit the use of PDSE.